A Bio-Inspired Slip Detection and Reflex-Like Suppression Method for Robotic Manipulators

Abstract

Robotic manipulators face significant challenges when handling objects of different sizes and shapes. Incorporating a sense of touch into these devices has the potential to improve performance and dexterity. In this paper, a bio-inspired approach is presented for slip detection and suppression during object manipulation. The method was inspired by the behavior of FA-I afferents located in the glabrous skin that encode sliding motion of objects over the skin. The proposed slip detection method encodes object motion captured by a slip sensor into spikes, following principles of neuromorphic sensing. The spikes are used as the feedback signal for an event-based closed-loop control system. The controller behaves in a reflex-like manner and actively engages the robotic fingers to increase grip force and suppress slip. A Dynamic Adaptive Threshold method was designed to improve slip detection for different surface properties of grasped objects. The performance of the method was evaluated following situations of dynamic slip caused by a sudden or gradual increase in object weight. The results demonstrated the feasibility of the proposed method. Slip events were suppressed before complete object slippage in 80% of all experimental trials. The response time ( $\Delta {\text {t}} ms) was compatible with the time for grip force adjustments in humans. This paper explored event-based touch applicable to the problem of manipulation, which is less explored than event-based tactile perception and shows promising prospects for both robotics and prosthetics.